Prognostic Value of Performance Status, Albumin, and CRP in Last-Line Chemotherapy for Pancreatic vs. Other Gastrointestinal Cancers-Simple Tools Matter

Abstract

Patients with advanced gastrointestinal cancers often receive chemotherapy near the end of life (EoL), raising concerns about overtreatment. The PALLiON trial, a cluster-randomized trial, assessed the impact of a complex intervention on frequency of EoL treatment; the intervention involved palliative care referrals and the use of PROMs. The present secondary analysis evaluated the prognostic value of baseline performance status (PS), albumin (alb), C-reactive protein (CRP), and body mass index (BMI) for overall survival, comparing pancreatic (PAN, n = 189) vs. other gastrointestinal cancer patients (GI, n = 286). Baseline PS, alb, CRP, mGPS (modified Glasgow prognostic score), and BMI were analyzed using Cox regression. Adjusted for age, sex, and hospital size, PS ≥ 2 and alb < 35 g/L predicted shorter survival in both PAN and GI cancers, while CRP > 10 predicted shorter survival only in GI cancers. In PAN, PS ≥ 2 predicted a 78.4% higher probability of shorter survival, and mGPS 2 predicted a 68.7% higher probability. In GI, mGPS 2 predicted a 70.8% higher probability, whereas PS was not significant. BMI did not improve predictive models. PS ≥ 2 and low albumin are strong predictors of short survival in PAN, whereas increased CRP and low albumin (mGPS 2) are predictors in GI.

Keywords: end-of-life chemotherapy; gastrointestinal cancer; palliative care; pancreatic adenocarcinoma.

Figure 1

The PALLiON trial assessed time-based vs. needs-based referrals to palliative care and use of PROMs, following patients from start of last-line chemotherapy until death. In the present study, we retrospectively compared PS, CRP, mGPS (modified GPS), and BMI as independent predictors of overall survival in advanced pancreatic cancer and other gastrointestinal cancers.

Figure 2

Cox-adjusted survival curves from start of last-line chemotherapy until death by baseline performance status (PS) and albumin (alb, ≥35 g/L vs. <35 g/L) among patients with pancreatic cancer (PAN; n = 189) and other gastrointestinal (GI) cancers (n = 289). (A) PAN, PS ≥ 2 vs. 0–1; p < 0.001. (B) GI, PS ≥ 2 vs. 0–1; p < 0.001. (C) PAN, alb ≥ 35 g/L vs. <35 g/L; p = 0.003. (D) GI, alb ≥ 35 g/L vs. <35 g/L; p = 0.003. Model adjusted for age, sex, and hospital catchment area, and additionally for PS at baseline (two categories, (C)), and albumin (two categories, (D)).

Figure 3

Cox-adjusted survival curves from start of last-line chemotherapy until death by mGPS at baseline among patients with (A) pancreatic cancer (PAN; n = 107, p = 0.005) and (B) other gastrointestinal cancers (GI; n = 149, p = 0.005). Model adjusted for age, sex, hospital catchment area, performance status at baseline (2 categories, PS 2–3 vs. PS 0–1), and mGPS (0, CRP ≤ 10 [reference]; 1, CRP > 10; 2, CRP > 10 and alb < 35 g/L).